Roundtable interview with AAA RPG devs at PC Gamer: 'We're running at a f**king wall, and we're gonna crash'

[Hace El Oso]

But how will this change by switching engine? You have the same with any engine, including in-house built ones across projects, when it comes to packages or default renderers.

Engines didn’t used to be built by a single company with the express intention of lowering the burden of competence on developers and offering a slew of cookie-cutter shortcuts to cobbling together something customers will pay for. They were purpose-built (and the best still are) by the people using them to create the specific kind of game they want to make. It helped ensure that the people making the games were motivated, dedicated, and able to adapt and overcome. That always results in a more worthwhile result.

How/what will switching to Unreal, EA's/Ubisoft/303 (who've also announced they've switched to Unreal) fix when they're indistinguishable for the average user? An imaginary problem?

This isn’t about users, this is about putting companies in a position where high-turnover, low personnel investment doctrine is not enabled and rewarded.

 

[Orud]

Which are these 'best' engines that are purpose built? And how did a custom built game engine result in more passionate games?

This isn’t about users, this is about putting companies in a position where high-turnover, low personnel investment doctrine is not enabled and rewarded.

So support Unreal then, so smaller companies created by people interested in making unique games can make them? These high turnover companies are the ones that still have their custom in-house engines.

 

[Hace El Oso]

Which are these 'best' engines that are purpose built? And how did a custom built game engine result in more passionate games?

The GEM engine for RTS. The VBS/Real Virtuality engine for military sims. The Digital Nature engine for flight sims. Clausewitz for grand strategy games. I could go on. All built by and for the same people using them to make specifically the type of game they wanted to make. No shortcuts, no free texture libraries, no easily replaceable personnel because every new person has to be brought up to speed.

These high turnover companies are the ones that still have their custom in-house engines.

Nonsense, if CDPR had used Unreal for CP2077 you wouldn’t even make that statement. But they’re just the special case of engine use still catching up with company doctrine, a lag which caused them terrible problems in production. All AAA game companies are rushing to emulate the zero-investment, high-turnover model because it’s cheap and easy. CDPR just got the order wrong, dynamiting the company faster than usual.

 

[Azdul]

Which are these 'best' engines that are purpose built? And how did a custom built game engine result in more passionate games?

This isn’t about users, this is about putting companies in a position where high-turnover, low personnel investment doctrine is not enabled and rewarded.

So support Unreal then, so smaller companies created by people interested in making unique games can make them? These high turnover companies are the ones that still have their custom in-house engines.

Unreal is not for 'smaller companies making unique games'.

First - Unreal is used to satisfy Nintendo (and Sony) requirement of 'no open source please on our precious consoles'. There are perfectly good smaller LGPL engines which lose to Unreal / Unity for this very reason.

Second - Unreal license is only cheap only if you sell your soul to Epic and publish EGS exclusive. No royalties - 'do what you want with the source code' license was few million $$$ some time ago. I'm not sure about current license terms - it may have gotten cheaper.

Third - There are the stories how Unreal engine reached 20 fps rendering nothing (world with no objects) - on hardware that could run slightly older games (like Doom 3) at 60 fps. Before you start making a game - you already have relatively high hardware requirements.

Fourth - you either do things Unreal way - or you have to rewrite large part of the engine. Best example was the limit of 64 players on older Unreal versions.

 

[ghostlife]

Third - There are the stories how Unreal engine reached 20 fps rendering nothing (world with no objects) - on hardware that could run slightly older games (like Doom 3) at 60 fps. Before you start making a game - you already have relatively high hardware requirements.

You mean like a pentium 4 lmfao? Doom 3 came out in 2004 and 60 fps isn't very high

 

[HeyCheckOutHowHardICanPee]

I'm not sure about current license terms - it may have gotten cheaper.

5% royalties once you make 1 million bux off of your product, nothing before that
This has been the case since 2015.

I can shittalk a myriad of issues and shortcomings of Unreal but the sheer lack of barriers of entry to the source code and getting started out without having to worry about licensing makes the engine look like the sweetest dream possible compared to the unholy depression hell that is the "Unity Source Code Negotiations".

 

[Orud]

The GEM engine for RTS. The VBS/Real Virtuality engine for military sims. The Digital Nature engine for flight sims. Clausewitz for grand strategy games. I could go on. All built by and for the same people using them to make specifically the type of game they wanted to make. No shortcuts, no free texture libraries, no easily replaceable personnel because every new person has to be brought up to speed.

Don't you think it's odd how all those engines you summarized are tightly coupled to a well-off game series, where each game is an iteration on the previous one? How the few non-main line games are basically reskins of existing ones? Nevermind that the companies maintaining the more complex engines are not small (+100-500 people)?

They keep upgrading these engines because its cheaper than porting all their tools and decade old game specific features to a new system. And that's the essential thing you're mixing up. The engine is basically a complex piece of paper that rarely does something technically extraordinary. The look, feel and content, basically everything that you would call a "game", is detached from that. It's up to the designers and artist that draw on that piece of paper via tooling, not by manually writing code.

Yet that bland, white piece of paper is so expensive (in time and money) to create and maintain. That's why you see the new projects (that differ from the specific game series) of the developers that you yourself linked are also switching to Unity or Unreal.

Nonsense, if CDPR had used Unreal for CP2077 you wouldn’t even make that statement. But they’re just the special case of engine use still catching up with company doctrine, a lag which caused them terrible problems in production. All AAA game companies are rushing to emulate the zero-investment, high-turnover model because it’s cheap and easy. CDPR just got the order wrong, dynamiting the company faster than usual.

Why wouldn't I have made that statement? Unreal is used by both big and small companies now, it has nothing intrinsically to do with the shitty companies that uses them. It simply enables anyone to do a lot more in less time than was usually possible.

AAA studio's are swapping to these prebuild engines because it offloads a lot of unnecessary research and development costs, in addition to their workforce being able to use the tools and tricks they learned at school or at home. Do you think it was only their management wanting to make the switch?

• CD Project Red's developers have stated multiple times over the years in leaked documents that they wanted to switch to Unreal for familiarity and similar features having to be custom built in their engine sometimes taking up months if not years.
• Read Bioware's post mortem about all their Frostbite (EA's in-house engine) woes and how the developers keep stating that they simply want to use Unreal again.
• Respawn Entertainment specifically made clear to EA that they do not wish to use any in-house engine like Frostbite. EA relented after almost a decade of Frostbite headache.
• etc... .

 

[Ravielsk]

You mean like a pentium 4 lmfao? Doom 3 came out in 2004 and 60 fps isn't very high

My guess is that if anything the problem was a fundamental incompatibility between the engine and the machine. DOOM 3 is a 32-bit app meant to run on Windows XP, Unreal 4 is entirely 64bit focused and barely meant to work on Win7.

That said it is a very poorly performing engine. Even in the early days I remember playing fan made games in UE3 that could not hit more than 20fps on my machine that was running Batman Arkham Asylum at 60fps. In fact on consoles UE3 was almost always a guarantee of sub 30fps. Unreal 4 was no better and I suspect that EU5 is much the same. Without extensive tweaking it just runs rather poorly.

 

[Orud]

Unreal is not for 'smaller companies making unique games'.

What? It absolutely is. Outside of its engine, Unreal comes with a very powerful editor, is well documented and all that is free of charge. Plus it has a very big community behind it to help anyone out.
They're even specifically investing in expanding/investing in their easy-to-use blueprint system that focuses on making scripts via a visual interface and basically no more code.

First - Unreal is used to satisfy Nintendo (and Sony) requirement of 'no open source please on our precious consoles'. There are perfectly good smaller LGPL engines which lose to Unreal / Unity for this very reason.

Could you link to an article or statement about this? It seems to me certain open-source software might be blocked because it's designed in such a way that they become attack vectors for pirates. In the recent past Nintendo has removed a certain Switch game because of exactly that.

Second - Unreal license is only cheap only if you sell your soul to Epic and publish EGS exclusive. No royalties - 'do what you want with the source code' license was few million $$$ some time ago. I'm not sure about current license terms - it may have gotten cheaper.

Both statements are false. As said by someone else, the licensing fees are 5% once your title earns 1 million. EGS exclusivity is by your own choice when they offer it, most simply take it because it's very lucrative with cash up-front. If any game studio is claiming otherwise, it's to try and hide their greed (or save face) by blaming Epic.

Third - There are the stories how Unreal engine reached 20 fps rendering nothing (world with no objects) - on hardware that could run slightly older games (like Doom 3) at 60 fps. Before you start making a game - you already have relatively high hardware requirements.

There are Unreal 4-5 engine games of which the minimum system requirements are 10 year old PC's. You need to keep in mind that it's not simply the speed of the hardware that counts. For such old systems, incompatibilities with modern drivers (requiring computing intensive adaption layers) alone can cause severe performance issues.
If you're really aiming for such low system requirements, something custom might be better to use. What happens more often though, is that another pre-built engine is used, like with OpenMW (which, by the way, can't run on the same hardware as OG Morrowind). You pick the solution that fits your requirements.

Fourth - you either do things Unreal way - or you have to rewrite large part of the engine. Best example was the limit of 64 players on older Unreal versions.

Yeah that's... true? Every version of Unreal (and engine, really) had its limitations. Earlier version of Unreal were not that great to create seamless open world games out of the box. That's why you have companies like Bethesda using Gamebryo for TES or Bioware Austin using the Hero Engine instead for their MMO because the Hero Engine comes with it's own server tech (which is handy if you want to create an MMO).

But the most important thing is that Unreal allows you to extend the codebase if you do want to build these features yourself. Plenty of Korean MMO's have decided to do the work and rewrite parts of the network architecture to fit in MMO's. You don't even need to use the Unreal editor if you really want, just keep in mind that it'll be labor intensive to set something different up. Do you know why Adobe Flash stuck around for so long? Because it's editor was amazing and people loved it (and still do to this day, last time I checked).

 

[Azdul]

First - Unreal is used to satisfy Nintendo (and Sony) requirement of 'no open source please on our precious consoles'. There are perfectly good smaller LGPL engines which lose to Unreal / Unity for this very reason.

Could you link to an article or statement about this? It seems to me certain open-source software might be blocked because it's designed in such a way that they become attack vectors for pirates. In the recent past Nintendo has removed a certain Switch game because of exactly that.

Blade Runner is the most recent example. Instead of using ScummVM source - Nightdive rewritten all the code by themselves - introducing plenty of bugs.

You either break GPL by not publishing the code - or Nintendo guidelines by publishing the code that contains calls to Nintendo API.

 

[deuxhero]

Case in point, there was a ScummVM port to Wii before the publisher realized it would need to release Nintendo's code under the license.

 

[Orud]

Blade Runner is the most recent example. Instead of using ScummVM source - Nightdive rewritten all the code by themselves - introducing plenty of bugs.

You either break GPL by not publishing the code - or Nintendo guidelines by publishing the code that contains calls to Nintendo API.

Ok, since you didn't want to provide a source yourself I did some searching of myself and found : https://www.eurogamer.net/replicati...dventure-game-classic-is-so-tough-to-remaster

Digging into the detail, if Nightdive had used ScummVM's software using the General Public License (GPL), it would have to supply that code and any modifications it made to it to the end user upon request. This is why every ScummVM game comes with the code to ScummVM, as well as anything else that may have come from the team.

But you can't do this on console because GPL-licensed code may not be legally compatible with the rules console makers have for publishing games on their consoles. Nightdive cannot, for example, sell somebody a game on Nintendo Switch and also provide them with accessible code upon request.

That has nothing to do with Nintendo being against ScummVM or any other open-source software. The problem is purely the GLP license, because it would require the publication of code that interacts with the system API that Nintendo/Sony/Microsoft keep locked behind a NDA.

ScummVM apparently works with a module-system to interact with various system API's, with a separate module for each system. So I can only guess that talks fell through because ScummVM required that any new module would also have to follow the GLP license.
Unreal, and many open-source software, circumvent this by separating their system specific modules from the rest of the public codebase. Usually under a separate license and under lock and key.

My guess is that if anything the problem was a fundamental incompatibility between the engine and the machine. DOOM 3 is a 32-bit app meant to run on Windows XP, Unreal 4 is entirely 64bit focused and barely meant to work on Win7.

That said it is a very poorly performing engine. Even in the early days I remember playing fan made games in UE3 that could not hit more than 20fps on my machine that was running Batman Arkham Asylum at 60fps. In fact on consoles UE3 was almost always a guarantee of sub 30fps. Unreal 4 was no better and I suspect that EU5 is much the same. Without extensive tweaking it just runs rather poorly.

Unreal isn't to blame for that, same with Unity. Too many people simply attribute hobbyist with too much skill, including the hobbyists themselves.
I can't count the amount of hobby projects I've seen of people that will tell everyone that they've worked for Microsoft or Google, yet they make sure the non-interactable carrot (taking up 1 pixel on the screen) has 500 000 polygons.

 

[Monocause]

I'm hearing this argument now for at least 10 years, and yet devs still keep pressing themselves into making bigger and unwieldier games that contain thousands of bugs at launch. Why?

The answer's pretty simple, actually - this argument was always correct but was offset by market growth. Up to a certain point you could offset growing the budget by pushing more sales, and pushing more sales became an option as going multiplatform got easier and the PC market got dominated by big digital distributors.

Problem is I don't think there's much growth to be had anymore and you can't offset the growing risk anymore. So we're reaching a point where one way or another you want to start maintaining the sales while cutting the costs.

It's the growing pains of an industry in transition. SaaS will be the future of the gaming industry, not because it is good, but because it's the model that's taking hold right now in the software development world. The most obvious element would be the cloud system, like Game Pass or the Play Station one. They aren't unpopular, but at the same time, console gamers are used to having to pay for multiplayer features. In fact, console gamers are used to pay for things that pc users (not just gamers) haven't thought of in their lives. I remember being very surprised when I learned that fact, like, why would you pay for access to the multiplayer portion of the game if you're already paying your ISP for your internet connection?
Regardless, SaaS also allows for constant monetization of a service. You're paying once for a video game. And you're paying 60 bucks. So a SaaS model allows you to pay 4,99 (for instance) per month so you can play a 60 bucks game. From a consumer standpoint, getting to play 5 AAA games for 5 bucks is a win win scenario, but then you have to consider other factors like the quality of your internet connection, which should only be a worry if you don't have access to fiber optic stuff or live in a country that cucks paying customers, like the First World countries (and in particular the USA), and then other, more nebulous aspects like ownership or even the ammount of money that studios are receiving per "stream". Streaming has been another way for higher ups to fuck the creator before they get fucked by them. However I wouldn't doubt that someone has already spoken of a Steam Pass during one of Valve's meetings.

Yes and no. SaaS is an option for certain kinds of products, or (in the case of straight-to-gamepass) products with a limited budget. And straight-to-gamepass development is a niche that exists mainly because owners of GoD platforms need to expand their library and boost subscriptions. Development like that isn't really profitable in most cases for the publisher.

The reason SaaS is taking hold in a lot of places in general software is that the software is used continuously (Office 365, Adobe CC etc). Thus SaaS is an evolution of an already existing practice with licensing/updates. In gaming you have the unique situation that doesn't exist in a lot of software where the grand majority of products sold are effectively one-off consumables. Even with non-MMO games that are praised for their replayability, it takes a special kind of person to keep playing it regularly for a long period of time and they're overall a minority in the gamer demographic. And for the games with extreme replayability there's other models, like Paradox-like focus on constantly releasing new DLCs and hence keeping the product fresh and generating revenue.

The SaaS model just wouldn't work for the grand majority of titles unless you significantly overhauled and MMO-ify them, and there's limits to what you can do there to keep a product being appealing. You can throw in online features into what's basically a single player game but in most cases it won't affect the big picture much.

tl;dr - SaaS works for MMOs (of which the ones with subscription fees have always been historically SaaS) and games which have multiplayer as the core feature. For single player games I don't see it happening, and game passes ultimately will function as a repository of older games that had the sales tail drop below the point of profitability. So first you go full-price distribution, then start dropping the price and doing promos and then sell your title to a GoD (games on demand) platform. The biggest change will be I believe that the GoD platforms will probably kill of the practice of game bundles as previously this was your last resort of squeezing extra sales.

 

[gurugeorge]

They've made a rod for their own backs, because the larger audience they started chasing some time in the 90s wants the cinematic storytelling and is less interested in the game side. But it's a trap, because it needs exponentially more time and effort, as they say, to make that kind of game, so the game has to attract that bigger audience justify the bigger budget. If it isn't a gigantic hit, if it's only a modest hit, it's a failure.

If they'd stuck with their natural audience and more modest budgets and returns, they could have continued as they had been going up to about the introduction of the PS2/XBox, and improved both gameplay and graphics, with just enough cinematics here and there to give a bit of pizzazz.

 

[GhostCow]

Games were better when they were games. I'll never understand why these fucktards in gamedev think they need to make their games cinematic. I fucking hate movies. I read their explanation and it's still retarded and makes no sense. They are entirely different mediums with different strengths and weakness.

 

[AwesomeButton]

People always tend to ignore the whole FMV genre

yea the mid 90's that was a bad time. As soon as the CD-ROM came along, the big studios wasted no time hogging up all that space with b-list community theater actor cut scenes

Not just b list, one of them had John Hurt and everything. But yeah, that was the most aggressive "games are movies" moment we've ever seen.

The one with John Hurt featured some quality "90s porn" (not with Hurt himself thankfully).

 

[Cat Headed Eagle]

I'm hearing this argument now for at least 10 years, and yet devs still keep pressing themselves into making bigger and unwieldier games that contain thousands of bugs at launch. Why?

The answer's pretty simple, actually - this argument was always correct but was offset by market growth. Up to a certain point you could offset growing the budget by pushing more sales, and pushing more sales became an option as going multiplatform got easier and the PC market got dominated by big digital distributors.

Problem is I don't think there's much growth to be had anymore and you can't offset the growing risk anymore. So we're reaching a point where one way or another you want to start maintaining the sales while cutting the costs.

It's the growing pains of an industry in transition. SaaS will be the future of the gaming industry, not because it is good, but because it's the model that's taking hold right now in the software development world. The most obvious element would be the cloud system, like Game Pass or the Play Station one. They aren't unpopular, but at the same time, console gamers are used to having to pay for multiplayer features. In fact, console gamers are used to pay for things that pc users (not just gamers) haven't thought of in their lives. I remember being very surprised when I learned that fact, like, why would you pay for access to the multiplayer portion of the game if you're already paying your ISP for your internet connection?
Regardless, SaaS also allows for constant monetization of a service. You're paying once for a video game. And you're paying 60 bucks. So a SaaS model allows you to pay 4,99 (for instance) per month so you can play a 60 bucks game. From a consumer standpoint, getting to play 5 AAA games for 5 bucks is a win win scenario, but then you have to consider other factors like the quality of your internet connection, which should only be a worry if you don't have access to fiber optic stuff or live in a country that cucks paying customers, like the First World countries (and in particular the USA), and then other, more nebulous aspects like ownership or even the ammount of money that studios are receiving per "stream". Streaming has been another way for higher ups to fuck the creator before they get fucked by them. However I wouldn't doubt that someone has already spoken of a Steam Pass during one of Valve's meetings.

Yes and no. SaaS is an option for certain kinds of products, or (in the case of straight-to-gamepass) products with a limited budget. And straight-to-gamepass development is a niche that exists mainly because owners of GoD platforms need to expand their library and boost subscriptions. Development like that isn't really profitable in most cases for the publisher.

The reason SaaS is taking hold in a lot of places in general software is that the software is used continuously (Office 365, Adobe CC etc). Thus SaaS is an evolution of an already existing practice with licensing/updates. In gaming you have the unique situation that doesn't exist in a lot of software where the grand majority of products sold are effectively one-off consumables. Even with non-MMO games that are praised for their replayability, it takes a special kind of person to keep playing it regularly for a long period of time and they're overall a minority in the gamer demographic. And for the games with extreme replayability there's other models, like Paradox-like focus on constantly releasing new DLCs and hence keeping the product fresh and generating revenue.

The SaaS model just wouldn't work for the grand majority of titles unless you significantly overhauled and MMO-ify them, and there's limits to what you can do there to keep a product being appealing. You can throw in online features into what's basically a single player game but in most cases it won't affect the big picture much.

tl;dr - SaaS works for MMOs (of which the ones with subscription fees have always been historically SaaS) and games which have multiplayer as the core feature. For single player games I don't see it happening, and game passes ultimately will function as a repository of older games that had the sales tail drop below the point of profitability. So first you go full-price distribution, then start dropping the price and doing promos and then sell your title to a GoD (games on demand) platform. The biggest change will be I believe that the GoD platforms will probably kill of the practice of game bundles as previously this was your last resort of squeezing extra sales.

You are thinking people play one game. We have long reached the point where people are playing every game that comes out and when they aren't buying them, they are pirating them. Steam solved an issue when people still bought games individually, but now that's starting to change too. Even single player games have Season Passes now. Pathfinder 2 had TWO season passes. The next age of gaming will be one that will see the hobby becoming one more streaming landscape alongside Spotify and Netflix. Because games *are* being used continously now. With a monthly fee of 4.99, you will be able to cover the neet audience, even.

 

[Nifft Batuff]

Imagine thinking the Sistine Chapel ceiling was meant to challenge the people looking at it.

It was.

 

[Monocause]

You are thinking people play one game. We have long reached the point where people are playing every game that comes out and when they aren't buying them, they are pirating them. Steam solved an issue when people still bought games individually, but now that's starting to change too. Even single player games have Season Passes now. Pathfinder 2 had TWO season passes.

- No, people aren't playing every game that comes out. Not sure where did you get that from.
- No, I don't think people play one game. I think - and can prove this with numerous data points - that people play a game and then stop playing it, and most of them never ever revisit the title.
- No, most single player games don't have season passes and never will. Most games actually never even have DLCs released.
- Season passes have nothing in common with the SaaS model. It's more akin to crowdfunding, where you're basically saying "pay us now and get stuff later". With all the season passes on the market you can still purchase DLCs/expansion packs individually after they're released.
- Not sure what do you mean about Steam solving something

 

[Stavrophore]

Engines are not cheap, but remember the GSC, they had like 30-70 devs and they've done quite a good job creating the graphical engine with poor's man Todd Howard radiant AI. I can't imagine studios like ubisoft having trouble to make engine every so often that are not soulless copy cat of their previous work with rudimentary features that create simplistic game.

 

[Orud]

Engines are not cheap, but remember the GSC, they had like 30-70 devs and they've done quite a good job creating the graphical engine with poor's man Todd Howard radiant AI. I can't imagine studios like ubisoft having trouble to make engine every so often that are not soulless copy cat of their previous work with rudimentary features that create simplistic game.

That was 20 years ago. Crafting a modern engine requires a lot more effort, something you'd think would be quite obvious.

They even tried it again with Cossacks 3 in 2016, which was such a huge piece of buggy shit that for their latest project they've swapped to the Unreal engine.

 

[Curious_Tongue]

Trannies, Niggers and Women write spaghetti code, are lazy, uninspired bunch that look at work environment the same way cat looks at swimming ppol. And no amount of ESG free money, midwit intellectual gymnastics and political corectness can change that fact. What's the reason for shitshow in AAA gaming industry? That's the reason. That's The Reason. THAT'S THE REASON FFS!
On the other hand: with AI tools and Steam as marketing and selling platform, look me make Elder Scrolls 6 before Bethesda

Someone found the source code for Watch Dogs: Legion and said the entire game was written in if/then statements

 

[Cat Headed Eagle]

Engines are not cheap, but remember the GSC, they had like 30-70 devs and they've done quite a good job creating the graphical engine with poor's man Todd Howard radiant AI. I can't imagine studios like ubisoft having trouble to make engine every so often that are not soulless copy cat of their previous work with rudimentary features that create simplistic game.

I'm sorry, are you bringing out Stalker's engine as a good example? The engine that was infamous for being so absolutely buggy and broken that nobody knows for sure how many bugs are there even after official and fanmade patches? Quite ironi for a game dealing with radiation mutations tbh.

 

[thesecret1]

Engines are not cheap, but remember the GSC, they had like 30-70 devs and they've done quite a good job creating the graphical engine with poor's man Todd Howard radiant AI. I can't imagine studios like ubisoft having trouble to make engine every so often that are not soulless copy cat of their previous work with rudimentary features that create simplistic game.

The cost of developing the engine is just one aspect of why so few companies do it. Aside from the initial cost you need to front, you also need to keep up with the competition (especially in AAA), meaning you will need a solid team working on it basically indefinitely. Furthermore, there is a "hidden" cost in that nobody outside your company knows how to use that engine – while you've got droves of devs that have experience with, say, Unreal or Unity thanks to them being used across many studios, you will have to train any new employee from the ground up with your engine. That's not to mention that while Unreal and Unity both have shit documentation (or rather, they DON'T have documentation for half their fucking engines), they also have shitloads of people around the world troubleshooting various issues on forums and elsewhere, so if you get stuck on some bullshit, chances are you can google a solution. That's not a possibility with an in-house engine because, again, nobody outside your company knows it.

That's not to mention that should your engine dev team fail for some reason (such as entering development hell because their codebase is shit), you face the decision of either continuing to foot the costs, being left behind your competition, while hoping your team gets its shit together soon, or axing it and switching to a different engine, thus losing your investment.

 

[Bad Sector]

That's not to mention that while Unreal and Unity both have shit documentation (or rather, they DON'T have documentation for half their fucking engines), they also have shitloads of people around the world troubleshooting various issues on forums and elsewhere, so if you get stuck on some bullshit, chances are you can google a solution.

That is what i thought before i worked on an UE game, but it turned out 99% of those forums are outdated, wrong and/or misleading and the best course of action is to just read the engine's source code myself.

Which is basically the exact thing i was doing when i worked at companies with custom engines.